Exploration and assessment of abnormal values ​​in some nonlinear models

Abstract

Was the presence of abnormal values ​​among a group of data significantly affects the results of statistical analysis of the data, from here show the importance of diagnosis of abnormal values ​​and mission as a first step in the process of analysis and decision making. Therefore, the diagnosis of homosexuals is one of the general goals in the process of data analysis, and estimate the values ​​are considered abnormal after diagnosis to end the Hzazha back to the alternative data values ​​and mission as a second step in the process of data analysis for the application of appropriate statistical analysis.Can summarize the aim of the research in two points, the first includes a study of some regression models non-linearity and the effect of its presence on the capabilities of these models, the second proposal of new methods for estimating Views anomalies in some models, nonlinear with comparisons between the estimation methods proposed and existing roads actually using the average of squares error MSE and the coefficient of determination.